So technologies can help us predict deadly risk factors

When I die, I want to be there too, Guareschi said. In the short term, perhaps, we may be present not so much at our death, but at its simple prediction, with a degree of accuracy that equals the most refined and complex predictors of mortality risk. The data of just six minutes of walking, collected through motion sensors of a common smartphone, could in fact be enough to predict the risk of death at five years. A group of researchers looked at data from 100,655 participants in the UK Biobank study, which collects information on the health of middle-aged and older adults who have lived in the UK for over 15 years. As part of a dedicated study, participants wore motion sensors on their wrists for a week. About 2% of the participants died in the next five years.

The researchers specifically focused their attention on about one-tenth of the participants, for whom movement data and five-year mortality data were collected. Using a machine learning model, an algorithm was then developed that estimated the five-year mortality risk using acceleration during a six-minute walk. It must be borne in mind that for many diseases, particularly in adults and the elderly and particularly for heart and lung conditions, a walk characterized by slowing down due to shortness of breath and short successive accelerations, with a very characteristic pattern of movement, has been demonstrated..

Also for some neurological conditions, such as Parkinson’s disease, a motor disturbance that can be monitored by sensors is showing in preclinical phases, and it is possible that it is possible to establish its predictive value. Once an algorithm was obtained that correlated the motor characteristics prevalent in an individual with the risk of death at five years in the initially selected subgroup, the model obtained was tested using the data of all the other participants. In particular, a quantity was calculated, called index C, used in biostatistics to evaluate the predictive capacity of risk by a model based on selected indicators.. The value obtained, equal to 0.72, is comparable to that obtained from other life expectancy estimation metrics, such as daily physical activity or health risk obtained through specific questionnaires.

Basically, a very simple predictor, tied to the monitoring of just six minutes of walking, has the same ability to measure much more complicated to obtain. While the latest published study used wrist motion sensors, mobile phones are now being used, capable of measuring acceleration and other walking characteristics, in order to take advantage of a potential sensor that is universally widespread and does not require investment and additional costs. Precisely for this reason, despite using more sophisticated motion sensors, the researchers used only one type of data that is commonly provided by cell phone accelerometers, in order to validate the use of the type of data obtainable from mobile phones to power the prediction algorithm they obtained.

The transition to direct measurement from a mobile phone, scheduled for a forthcoming large study, is therefore ready; we will see if the different way of wearing the sensors will have any influence and even if there are subpopulations of greater interest – for example the elderly in a certain age group, or those suffering from cardiopulmonary conditions – for which it is possible to obtain even more estimates accurate.

Why Predict Your Risk of Death?

The hope is that, as always when it comes to risk monitoring, there is the possibility of intervening in a targeted manner on those who show worsening or who appear to be at more serious risk, directing particular care in the light of an easily obtainable risk estimate. Of course, however, the market appetite for such mortality predictors must be taken into account, which can introduce discrimination – think for example of insurance companies and employers who had access to such data. It is therefore essential that the concrete applications of this type of research, further ahead than we imagine, be subjected to scrutiny by the regulator, using the criteria already adopted for other commonly used risk predictors (for example the famous insurance questionnaires); in fact, abuse can only be prevented by far-sighted legislation, which exploits the advantages obtainable from technology without forgetting the risks.